APPLICANT'S ABSTRACT: The aim of the project is to determine the effect of changes in the minimum legal drinking age (MLDA) on rates of youth (a) homicide, (b) aggravated assault, (c) suicide and (d) drowning. Injury and crime are among the more important health and social problems of American adolescents. Changes in the MLDA in many states in the 1970s and 1980s created a "natural experiment." Dozens of studies have explored the impact of MLDA changes on drunk driving and traffic injuries. Very few have examine the effect on other alcohol-related youth problems. Simple before-and-after state-by-state comparisons will be made, but the primary model is an econometric "fixed effects" model, using pooled time- series and cross-sectional state data for the 48 contiguous states from 1970- 1990 (about 1,000 observations). The dependent variable is the log of the particular youth injury or youth violence rate. Data come from the National Center for Health Statistics death certificate tapes and from the Uniform Crime Reports. The key independent variable is a dummy variable for the MLDA for the specific state-year. Other dummy variables for the "state" effect and the "year" effect are included as controls. Separate regressions will also be run by race and sex. MLDA enforcement proxies (e.g. alcohol arrest rates) will be added in subsequent runs. To ensure robustness of results, a second econometric model will be employed for homicides and aggravated assault. State and year dummy variables will be replaced by specific factors which are expected to affect these crimes. The theoretical framework is the economic model of rational behavior. Since crime and law enforcement are interrelated, a simultaneous equation will be used. The equation will be estimated in log linear form, and the coefficients estimated using two-state weighted least squares. The law enforcement variable is the arrest rate. Independent variables in the model include proxies for alcohol prices and narcotic consumption as well as education, urbanization, migration, divorce and strictness of handgun control laws. Most observational studies can determine only associations, not causation, for associations can be explained in other ways. By contrast, analysis of the MLDA natural experiment can indicate whether an intervention designed to affect alcohol availability among youth also affects crime and drowning.